履带式收割机全田块路径跟踪算法设计与试验

针对中小型传统收割机无人化作业需求，该研究设计了一套后装式全田块自动驾驶系统。以沃得锐龙单边制动转向型履带式收割机为平台，搭建手自兼容自动驾驶系统，并进行系统特性辨识试验，明确其测控性能限制。针对测控性能限制设计一种PD-Fuzzy-BangBang组合路径跟踪算法，并进行样机集成与试验。水泥地面直线行驶试验表明，组合算法相较单一PD算法的上线距离缩短57.3%，稳态标准差缩小81.3%。全田块模拟试验证明，组合算法在理想条件下的路径跟踪最大偏差为6.00 cm，标准差为2.42 cm，样机具备全田块自动驾驶功能。实际水田收割作业试验证明，样机在车速0.7 m/s条件下，上线过程的路径跟踪最大偏差为12.00 cm，标准差为 6.18 cm，全田块不漏割，割幅利用率大于80%，满足田间作业需求。     

遗传算法及其在渔饲料配方中的应用

首先阐述了参数编码、初始群体、适应度函数、遗传操作以及算法控制参数等遗传算法基本原理,然后着重介绍其在渔饲料配方中的应用,在该应用中采用了实数编码、基于线性排名的选择、算术杂交和非一致变异等遗传算法方法。结果表明,在渔饲料配方优化中应用遗传算法要优于目前饲料配方软件中大多采用的常规应用数学的优化方法。主要体现在单位产量饲料配方中的成本价格前者要低于后者。由于遗传算法能够解决因子较多,非线性程度高的问题,从而得出的饲料配方能更好的符合相关鱼种的营养含量标准。     

关于无网格方法中点插值形函数的研究

点插值法与其他无网格方法不同的是采用多项式近似来构造形函数，这种形函数具有Kronecker delta函数的特性，因此，易于施加本质边界条件．本文研究了点插值法中以单项式为基函数的形函数的建立及其性质，并通过矩阵三角化算法来克服形函数矩阵大奇异性．同时，本文所给出的数值算例验证了形函数具有Kronecker delta函数的特性，说明了点插值形函数具有精确的曲线拟合特性并能通过分片试验．     

一类矩形切割的优化模型

研究了工业中常遇到的先将原料板切割成加工拼板后，再将拼板切割成单元板的一类矩形板材切割问题，此类问题归结为二维排布的双层优化，针对单元板在拼板上和拼板在原料板上的4种不同排布情形，建立了统一的非线性整数规划模型，根据问题的特点，给出了该模型的一个化双层优化为单层优化的求解算法，实际应用中，该算法能在数秒钟内按工艺要求给出最优切割方案，与传统方法相比，料板利用率常可提高5％～10％．     

基于核K—均值聚类算法的植物叶部病害识

针对植物叶部病害图像的特点,首先对采集到的玉米病害彩色图像采用矢量中值滤波法去除噪声,然后提取玉米病叶彩色图像的纹理特征和颜色特征作为特征向量,利用Mercer核,把输入空间的样本映射到高维特征空间进行K-均值聚类以及植物病害识别.试验涉及的4种玉米病害识别正确率达82.5%,核K-均值聚类方法适合玉米叶部病害分类.     

面向安全服务重组多目标优化的粒子群遗传算法

遗传算法具有快速随机的全局搜索能力,但局部搜索能力差,易陷入早熟收敛,迭代效率低.粒子群算法采用速度——位置模型,可以较快收敛到指定精度.将粒子群算法与遗传算法融合,采用多目标遗传算法得出初步的优化结果,并将其作为粒子,利用粒子群算法强化局部搜索,加快收敛速度,仿真结果证明了该算法的优越性.在CSSM对底层安全服务的重组时利用粒子群和遗传算法的结合(GAPSO),能够提高效率.     

汽车多路传输系统数据的简化算法

介绍了一种应用于汽车多路传输系统中所有关联数据库数据的简化算法.同时对此算法进行了具体的性能分析.汽车多路传输系统允许汽车电子系统内各个智能模块共享信息,实现了信息在各个电子模块中进行实时交换,从而能提高各个智能模块之间的信息交换速度.     

The method for designing the profile of forest roads supported by genetic algorithm

This study examines a method for designing an optimum profile of a forest road in a short matter of time. In this method, a genetic algorithm (GA) searches for points where the gradient changes and dynamic programing (DP) designs the longitudinal slope. This method can be applied to the ground heights of an actual forest road. The evaluation function of the DP assesses the total expenditure. The inverse of this function is the fitness function. Crossover and mutation are operators of an evolution simulation of a GA. They are each repeated 20 times and 420 profiles are designed and assessed, all within about 10 min. A comparison of the profile designed by the GA with that of an actual forest road shows only one section where the formation level differed substantially. Overall, both profiles seem to show a high degree of correspondence. From this result, we believe that the profile designed by the GA is best, or the one of the best. A conventional method takes about 65 h to calculate all changing points of gradients. However, by using the GA, the calculation time was reduced to about 1/389 that of the coventional method, showing that the method which uses the GA through the profile design is very useful.     

Effects of polyline simplification of dynamic GPS data under forest canopy on area and perimeter estimations

The objectives of this study were to determine Global Positioning System (GPS) positional errors while moving under the forest canopy and to clarify the effects of polyline simplification on area and perimeter estimations. We used the Pathfinder Pro XR and GPSMAP 76S, which are categorized as “high-end mapping” and “general navigation” GPS receivers, respectively. The field tests were conducted in both natural and plantation forests. The results showed that the Pathfinder Pro XR, which has better multipath rejection technology, worked well, especially in the plantation forest under unfavorable conditions of higher stand density. We used analysis of variance to clarify the effects of the receiver type, positioning mode, stand type, and polyline simplification method on area and perimeter estimations. The receiver type and positioning mode were found to be significant factors that affected area estimation. The Pathfinder Pro XR estimated the area more accurately than the GPSMAP 76S, and differential GPS estimated the area more accurately than autonomous GPS. With respect to the perimeter, the receiver type, positioning mode, and polyline simplification method were found to be significant factors. The results showed that perimeter estimation was improved by using the velocity filter, and further improved by using the velocity filter and Douglas-Peucker algorithm, especially when the Pathfinder Pro XR was used. The GPSMAP 76S estimated the perimeter accurately without any filtering because its default speed filter worked well, even though the GPSMAP 76S is a general navigation GPS receiver.     

实数型遗传算法的研究

在引入实数型遗传算法的基础上,对遗传算法的关键问题交叉和变异操作进行了深入的研究,提出了几种新颖而实用的交叉和突异操作方法。同时,对需编码的传统的二进制型遗传算法和无需编码的实数型遗传算法进行了简要的比较,阐述了实数型遗传算法的优越性。